metis-desk/.agents/skills/ux-inputs-and-forms/SKILL.md
David 261de87a1f Add UX skills for navigation, notifications, search, and tables
- Introduced `ux-navigation` skill for designing and critiquing navigation systems, emphasizing information architecture and wayfinding best practices.
- Added `ux-notifications-and-toasts` skill to guide the design of notifications, focusing on urgency, user attention, and opt-out controls.
- Created `ux-search` skill to enhance search experiences, covering input, suggestions, results, and zero-results recovery.
- Implemented `ux-tables` skill for designing data tables and lists, addressing alignment, interaction, and data presentation.
2026-10-21 21:55:06 +02:00

12 KiB
Raw Blame History

name description license metadata
ux-inputs-and-forms Design or critique forms and input fields — field selection, labels, layout, validation, error states, and submission. Applies the always-true core (minimal fields, one column, labels outside inputs, inline validation) and gates levers like multi-step splitting, floating labels, and smart defaults. Trigger when the user asks to design, build, or review a form, signup/checkout/survey fields, input validation, or error messages. MIT
author version
uxcel 1.0.0

Inputs & Forms Skill

How this skill behaves (read first)

This is a generative skill. The default failure modes: over-asking (every field that might be useful, all required), mislabeling (placeholders as labels, jargon, unmarked optionals), and validating late (errors dumped in a modal at submit). The fix is not "apply every form pattern" — helper text, tooltips, progress bars, and multi-step splitting everywhere produce a bloated form. So this skill gates:

  1. Establish context — what the form collects, why each field is needed, platform, and length.
  2. Apply the always-true core — minimal fields, scannable single-column layout, persistent labels, the right input type per field, inline validation.
  3. Surface the context-dependent decisions (multi-step, floating labels, helper text, pickers, smart defaults) with trade-offs.

Then it hands off to ux-heuristics-audit, ux-accessibility-audit, ux-aesthetics-audit, ux-microcopy-audit, and ux-dark-patterns-audit for validation.


Step 0 — Establish context before designing

Ask if not known; state the assumption if proceeding without an answer:

  • What does each field buy? For every field, name who uses the data and for what. A field with no answer gets cut — the single highest-leverage form decision.
  • Platform — mobile demands contextual keypads, ≥38px touch areas, a submit button that stays in sight while scrolling, one-handed reachability.
  • Length & stakes — a 3-field newsletter signup and a multi-page checkout need different structure (steps, progress, error summaries).
  • Sensitivity — phone numbers, payment, IDs: users ask "why do you need this?" — plan the justification (helper text) and expect dark-patterns scrutiny on anything beyond task needs.

The always-apply core (correct for almost every case)

Ask less

  • Remove every field the task doesn't require. Shorter forms complete more; the system shouldn't ask for what it can detect or already knows (autofill known data, support browser autocomplete, auto-detect country/locale).
  • Mark required and optional fields — asterisk (before the label in LTR; not pale gray) for required, "(optional)" in the label for optional.
    • ❌ Ten required fields, no explanation ✅ Five fields, one marked "(optional)", helper text on the sensitive one

Layout for scanning

  • One column, top to bottom. Multi-column forms cause skipped fields, wrong-field entries, and abandonment (Baymard). Natural reading flow wins.
  • Labels above inputs, outside the input, always visible. Never placeholder-as-label: it disappears on focus, strains memory, and looks like filled data. Placeholders only give example values, visibly muted.
  • Group related fields (personal / billing / shipping) with headings and spacing — proximity does the structure: label-to-its-input gap < input-to-next-field gap; spacing between groups > spacing within.
  • Visible input boundaries — borderless "minimal" inputs make users hunt for where to type. Input height ~32–40px, text ≥16px, labels ~13–14px, traditional restrained styling.

Pick the input type from the data

  • Single-line for short answers; multi-line/text area for prose; password with a show/hide toggle; search with auto-suggest; select/dropdown only for long lists — up to ~5 options use radio buttons (single choice) or checkboxes (multiple); input masks wherever format matters (phone, date, card) — accept any way users type and format it for them; contextual keypads on mobile (numeric pad for numbers); OTP inputs that support paste and auto-fill.
  • Date fields: match the picker to the task — text input + mask for known dates (birth date), calendar picker for near-future scheduling (booking).

Validate inline, kindly

  • Validate on blur, not on every keystroke and not at submit. Let users finish typing; check when focus leaves the field. (Exception: live requirement checklists for new passwords.)
  • Error messages sit at the erroneous input (right side or directly below — never only at the top), in plain, polite, specific language: what happened + how to fix it, ideally one line, never truncated. Color + icon + text — never color alone.
  • Keep the error state visible until the input is actually valid, not just until the user starts typing.
  • Prevent before you correct: constraints, forgiving formats, and masks beat error messages.

Submission is part of the form

  • Primary button always in sight (below the fields; fixed on long mobile forms), labeled with the action — finish the user's sentence "I want to…": ❌ "Submit" ✅ "Subscribe" / "Place order" / "Create account".
  • Button hierarchy: primary visually dominant; "Cancel"/"Back" subdued; enough spacing that the wrong button can't be hit by accident. Touch target ≥1×1cm (~38px+).
  • Show all three states: loading after submit, a clear success message, and a polite, informative failure message (mind security: don't reveal which credential was wrong).

The context-dependent decisions (surface, don't auto-apply)

Present each with its trade-off and a recommendation tied to Step 0. Applying all of these to a 4-field form is the bloat failure mode.

Decision Apply when Avoid / adapt when Default recommendation
Multi-step + progress tracker ≥3 logical steps (checkout, surveys); keeps each page digestible Short forms — splitting adds clicks; >5–6 steps overwhelms Single page until grouping alone stops working
Floating labels Space-tight layouts wanting label + example in one slot Complex/rarely-used forms where a permanent label is safer Static labels above inputs; floating labels as a space-saving variant
Helper text Sensitive fields ("why do you need my phone?"), nuanced requirements (password rules) Every field "just in case" — noise that buries the signal Only where users predictably hesitate; one line max
Icons / tooltips for guidance Visual questions (CVV location) where a picture beats words Tooltip-only help on mobile (extra tap, unreliable hover) Icon first; tooltip as a supplement, not the sole carrier
Auto-suggest / incremental search Long option lists (countries); bold the matched characters Short lists where radio/checkboxes are faster Auto-suggest for any list users would have to scroll
Smart defaults / auto-detection Geo-detect country, prefill known data — fewer actions Defaults that look like a choice the user made (consent, marketing opt-ins — dark-patterns territory) Prefill facts, never decisions
Single full-name field Reducing friction; more inclusive across cultures The product must reliably split first/last name One name field unless the data model truly needs two
Error summary at top Long forms where multiple errors may be off-screen; each item links to its field Short forms — inline messages suffice Inline always; summary added for long forms
CAPTCHA / verification Abuse is real; prefer accessible alternatives (magic links, OTP, passkeys) Distorted-text CAPTCHAs; blocking paste in password fields Risk-based, accessible verification; never block paste
Left-side labels Extremely long desktop forms where vertical space is scarce Mobile, mixed label lengths, anything responsive Labels above inputs

Validate the result (orchestration)

Hand-offs name each lens by its installable skill name. Invoke one only if that skill is installed; if it isn't, this skill's own core already carries these rules — proceed without it rather than blocking.

After generating or revising, hand the result to the audit lenses rather than declaring it done. These are candidate lenses — posture is set by docs/orchestration-policy.md, or route the whole thing through ux-design-review. Here, heuristics, microcopy, and dark-patterns are Tier A (auto-run); accessibility, aesthetics, and mobile-responsiveness are Tier B (offered) — under an existing design system aesthetics is suppressed and accessibility narrows to usage; mobile-responsiveness applies only on mobile. If the user invoked this skill for one specific thing, respect that scope.

  • ux-heuristics-audit (Tier A) — visibility of system status (states, loading, success), error prevention, user control, recognition over recall.
  • ux-accessibility-audit (Tier B — offer; narrow under a design system) — keyboard operability and focus indicators, label/contrast minimums, aria-describedby error linking, multi-cue error states, accessible authentication.
  • ux-aesthetics-audit (Tier B — offer; suppress under a design system) — spacing rhythm, grouping clarity, whether "minimal" styling destroyed input affordance.
  • ux-microcopy-audit (Tier A) — label clarity, helper-text brevity, error-message tone and specificity, action-button phrasing.
  • ux-dark-patterns-audit (Tier A) — over-collection of personal data, pre-checked opt-ins, sneaky defaults, required fields that should be optional.
  • ux-mobile-responsiveness-audit (Tier B — offer; applies only on mobile) — when the context is mobile: keypads, touch areas, fixed submit, one-handed reach.

If the audits surface a conflict (e.g., marketing wants more fields vs. completion rate), resolve back toward the primary task: the user came to finish something — every field and interruption must earn its cost.


Common do/don't patterns

❌ Don't ✅ Do
Ask for everything that might be useful someday Every field justified by who uses the data and why
Placeholder text as the only label Persistent label above the input; placeholder = example only
Two-column field grid to "save space" One column, logical groups, natural reading flow
Dropdown for a 3-option choice Radio buttons (single) / checkboxes (multiple) under ~5 options
Force a strict typing format ("no dashes!") Forgiving formats + input masks that format for the user
Validate on every keystroke, or only at submit Validate on blur; live checklist only for password creation
Error modal at the top listing field numbers Specific, polite message at the erroneous input (+ summary only on long forms)
Generic "Submit" on everything Contextual action labels: "Create account," "Place order"
Submit with no feedback Loading → success/failure states, always
Pre-checked marketing opt-in as a "smart default" Prefill facts, never decisions (→ dark-patterns)
Ship without checking Hand off to heuristics + accessibility + aesthetics + microcopy + dark-patterns

Source lessons (Uxcel)